The aim of this study was to establish whether long-term use of chlorhexidine would prevent skin colonization by antibiotic-resistant Staphylococcus epidermidis. Ten nurses, working on a ward for haematological disorders, volunteered to participate in the test. They washed one arm every morning for three weeks with chlorhexidine gluconate, ('Hibiscrub' ICI Pharmaceuticals). The other arm served as a negative control. Samples from the antecubital fossa of both arms were taken two to three times a week during the wash period and two weeks thereafter, giving a total of 216 samples. The appearance of resistant S. epidermidis with different antibiograms was analysed. During the wash period the total bacterial counts and the counts of the resistant S. epidermidis strains on the test arm were both about one-tenth of those on the control arm, a significant difference (P < 0.05). Moreover, there were significantly fewer resistant S. epidermidis on the test arm, 1.3 per sample, than on the control arm, 2.5 per sample (P < 0.01). Most of the resistant S. epidermidis were only found once or a few times on the same site, after which they disappeared, though a few persisted on the skin even during 'Hibiscrub' washing. In an agar dilution test, chlorhexidine minimum inhibitory concentrations (MICs) of persisting strains were the same as for strains disappearing from the skin following 'Hibiscrub' washing, 1.0 or 2.0 mg l-1, but somewhat higher than MICs of strains isolated from healthy carriers outside the hospital whose MICs were 0.5 mg l-1. The relative contribution to the skin counts by those S. epidermidis strains found only occasionally were compared with those found repeatedly but no difference in reduction was found between these categories during 'Hibiscrub' washing.